Hacker News new | ask | show | jobs
by ivansavz 5388 days ago
I watched the news conf. this morning.

The neutrino production "curve" looks like a a square wave pulse as a function of time

____| |_____

and the detection points look like this

___________________________________| |_____

with ∆t ~= 700km/c time distance between the two pulses.

The claim is that the "most likely guess" of the received pulse shape (obtained from many measurements) is too far right to be consistent with speed of light, but the ∆t measured where?

between the onset of the pulses? between the place where the pulses go down?

What they did some mega calculation (maximum likelihood stuff), to predict the best approximation to the shape of the pulse at the receiving end -- so somehow they take on the approximation of the whole pulse (which is much wider than the claimed discrepancy).

They should downgrade the claim from: "speed of neutrinos is...." to: "speed of pulses of neutrinos ..... on average, as predicted by the maximum likely shape of the pulse", which sounds much less profound.

The conf was good though, the speaker stood up to a lot of serious scrutiny. My guess is the problem is with the ML curve shape calculation.

3 comments

They should downgrade the claim from: "speed of neutrinos is...." to: "speed of pulses of neutrinos ..... on average, as predicted by the maximum likely shape of the pulse", which sounds much less profound.

This would apply to every paper in existence. If you want to know the methodology, you read that section of the paper. Putting it in the title only obscures the message.

It's true that it makes the analysis a lot harder, and one of the questions after the talk was about making shorter bunches of neutrinos. The duration of the pulses are probably just set by the width of the proton bunches in the ring, in which case it would be nontrivial to make it much shorter.

I've read the article[1] and what they do is compare the proton flux as measured in Geneva with the neutrino signal measured in Gran Sasso. The proton signal is timeshifted a small amount until the curves overlap maximally.

Where the leading and trailing edge are is irrelevant here, as the complete shape of the pulses are compared.

1: http://static.arxiv.org/pdf/1109.4897.pdf

If the pulse is wider than the error, this opens up the phase velocity / group velocity avenue to explain the paradox.